| Summary: | Water saturation (Sw) is one of the significant parameters in hydrocarbon volume estimation. However, accurate estimation of this parameter is always difficult due the presence of clay minerals in the formation, which has a direct impact not only on the well log but also the core analysis data. The excess conductivity of the clay minerals would diminish the resistivity log data in the hydrocarbon reservoir leading to overestimation of Sw. In the experimental works of core samples, the clay conductivity would influence the determination of cementation factor and saturation exponent. All the aforementioned factors together with porosity would compromise the accuracy in Sw determination if mathematical model in conventional technique is used. With the evolution of artificial intelligent (AI) in the oil and gas industry, many had benefited from this approach by implementing data prediction from a single input parameter. This would reduce the impact of uncertainties from several input parameters such as in the case of Sw determination. In this paper, a new technique for estimating Sw using AI and integrated petrophysical analysis is introduced. A program code written in Python was established using Extreme Gradient Boosting (XGBoost) method with resistivity index (RI) from core analysis as input for model training and well log resistivity data for Sw prediction. Performance prediction was evaluated using mean squared error (MSE), root mean square error (RMSE) and R2. © 2024 Author(s).
|
|---|